Synthesis of. alpha.-dehydrobiotin

Jul 14, 1975 - (7) S. M. Kupchan, J. G. Sweeny, R. L. Baxter, T. Murae, V. A. ... George F. Field,* William J. Zally, Leo H. Sternbach, and John F. Bl...
0 downloads 0 Views 694KB Size
J . Org. Chem., Vol. 41, No. 24, 1976 3853

Synthesis of a-Dehydrobiotin dosage range. (6) Piscicidal activity was assayed using a procedure similar to that described by W. A. Gersdorff, J. Am, Chem. SOC.,52, 3440 (1930). Gnidilatin, gnidilatidin, and gnidiglaucin showed toxicity at concentrations of 50 /.Lg/l. (7) S. M. Kupchan, J. G. Sweeny, R. L. Baxter, T. Murae, V. A. Zimmerly, and

B. R. Sickles, J. Am. Chem. SOC.,97, 672 (1975). (8) K. Sakata, K. Kawazu, and T. Mitsui, Agric. Biol. Chem., 35, 21 13 (1971). (9) A similar proposal has been advanced for the role of the long aliphatic chain in the piscicidal activity of huratoxin; cf. ref 8.

Synthesis of a-Dehydrobiotinl George F. Field,* William J. Zally, Leo H. Sternbach, and John F. Blount Chemical Research Department, Hoffmann-La Roche, Inc., Nutley, N e w Jersey 07110 Received March 5,1976 a-Dehydrobiotin has been synthesized from the bicyclic sulfonium salt 7. The five-carbon acid side chain was elaborated by cleavage of the sulfonium ring with acetate ion, hydrolysis to a hydroxypropyl side chain, oxidation to a propionaldehyde side chain, and coupling with triethyl phosphonoacetate. A new reagent, orthophosphoric acid, was used to effect debenzylation of the urea moiety.

a-Dehydrobiotin (1) is an extremely effective antagonist of biotin that has been isolated from a natural source as a 0

J

H

NH

1

consequence of its antibiotic activity against a variety of bacteria and fungi.2 Subsequently, it was reported that a dehydrobiotin is accompanied by two other biotin antagon i s t ~and ~ that it is a product of the catabolism of b i ~ t i n . ~ We undertook synthesis of this biologically interesting molecule since we had available experience and intermediates derivingfrom the synthesis of biotin i t ~ e l fThe . ~ approach we followed was to divert the biotin synthesis cited above at a stage in which the difficult stereochemical problems have been solved and in which a reactive center is present where the double bond is to be formed. Debenzylated Series. The intermediate which was expected to be the most useful in this kind of approach is the cyclic sulfonium salt 2;jc however, as will be seen, an unexpected difficulty developed. The first step is oxidation of the terminal carbon atom of the latent side chain in 2 to the oxidation level of an aldehyde. Analogy from the biotin synthesis5 suggests that sulfonium salt 2 can react as if it were a covalent bromide with the appropriate side chain. Accordingly 2c6 was reacted with the sodium salt of 2 - n i t r o ~ r o p a n eto , ~ give a compound assigned structure 3c, a hemiacetal form of desired aldehyde. Since a priori sulfonium salt 2 has two other points where attack of the nitronate anion could have occurred, hemiacetal 3c was submitted to x-ray crystallographic analysis for confirmation of its structure. Two stereoscopic views of the result are shown in Figure 1.Additional confirmation was obtained by reaction of 3c with methanol or aniline under acidic conditions to give 4c or 5c, respectively. Reaction with hydroxylamine gave oxime 6c, the only compound of the series with an actual rather than a latent side chain. The next operation to be carried out is addition of the remaining two carbon atoms of the side chain. Reaction of 3c with triethyl phosphonoacetates should have been feasible, but we could obtain no product. Perhaps the hemiacetal ring is so stable that there is no appreciable concentration of the aldehyde form. Reaction of 3c with malonic acid and piperi-

dine gave a product whose elemental analysis and mass spectrum are consistent with a dimer of a dehydration product of 3c. This propensity of 3c to self-condense was also evident when treatment of 3c with acetic anhydride gave a similar “dimer” acetate. Since these products did not appear to have any synthetic utility, their structures were not investigated further. The NMR spectrum of the “dimer” is not readily interpretable, but is clearly not consistent with any symmetrical dimer. Racemic Series (a).These unproductive results forced a retreat to the precursor of 3, the dibenzyl derivative 7 (X = Br). Here we chose a lengthier reaction sequence. The ring was cleaved with acetate ion to give 8a whose basic hydrolysis gave alcohol 9a. The next operation was to specifically oxidize the alcohol function to an aldehyde without affecting the thioether function. This was achieved with dimethyl sulfoxide/dicyScheme I 0

2

4

I

K PHNHJH

+

H N D N H c f i H 5

Dimer - H,O dimer acetate - 2H20

H 6

Field, Zally, Sternbach, and Blount

3854 J. Org. Chem., Vol. 41, No. 24,1976

Figure 1.

clohexylcarbodiimide combinationg in 60% yield. The twocarbon fragment was added with the sodium salt of triethyl phosphonoacetates to give l l a in 50% crude yield. There remained only removal of the benzyl protecting groups to complete the synthesis. Two reagents for debenzylation of the cyclic urea moiety are known. Neither one proved very satisfactory owing to the

a

C,HbCH,--N

Scheme I1

jV-CH&H,

X-

K

CcHjCHZ-N,

\

0 +si

Y--CH,C,Hj

I

Q R-

8, R = CH,OCOCH, 9, R = CH,OH

7

/

= CHo

J I

i

%'

+

0

13

HN

a

.

1

NH

,

0

HN

14

K NH

1

reactivity of the a,@-unsaturatedester functionality. Using one, sodium in liquid ammonia,5a on 1 la we obtained a reaction mixture whose NMR spectrum contained peaks for the vinyl hydrogens a t only about half the expected intensity. Presumably the conjugated double bond was partially reduced and consequently this method was not investigated further. Using the other, concentrated hydrobromic acid, we obtained, after brief heating, a product in 67% yield to which structure 13a was assigned on the basis of its NMR spectrum (no vinyl protons). Prolonged heating under reflux gave a poor yield of material presumed to be the hydrobromide of the debenzylated zwitterion 14. Since we feared a fragmentation reactionlo as indicated by the arrows on 14, this crude reaction product was treated directly with methanolic hydrogen chloride to esterify the carboxyl group. Treatment of this reaction mixture with sodium bicarbonate then gave a low yield of methyl ester 15 which on alkaline hydrolysis gave dl-a-dehydrobiotin (la). Optically Active Series. Synthesis of d-dehydrobiotin (1 b) followed along the same lines; that is, the sulfonium salt 7b (X = d-camphorsulfonate) was converted to aldehyde lob, the side chain was lengthened with triethyl phosphonoacetate, and then the benzyl groups were removed. However, some experimental modifications could be incorporated with profit. Since we had available large quantities of the optical antipode 7c (X- = d-camphorsulfonate) a good many model experiments were made with material derived from this substance. Optically active aldehyde 10b could not be obtained crystalline. On the suspicion that impurities associated with the dimethyl sulfoxide/carbodiimide method were preventing crystallization and easy purification, other methods of oxidation were examined. Both the Collins reagentll and pyridinium chlorochromate12 gave 10b in approximately 40%yield accompanied by some sulfoxide 7b. Thus both of these methods possess a good degree of specificity for oxidation of an alcohol group in the presence of a sulfide. Even though the aldehyde still remained an oil, a t least it was demonstrated that these three methods are about equivalent in this case. The problem of removing the benzyl groups of l l b was solved in a much more efficient manner than previously by using orthophosphoric acid containing phenol as a benzyl acceptor. The yield of l b from l l b was 53% without any indication of cyclized products such as 13 and 14 which had been obtained using hydrobromic acid. A possible explanation for these differing results might be that the sulfonium salts are formed only by displacement of bromide ion from an intermediary @-bromoacid derived by addition of hydrogen bromide to the double bond. I t does not appear that orthophsphoric acid has been recognized previously as a useful reagent for removing benzyl groups. However, polyphosphoric acid has been used for the removal of a-methylbenzyl groups from amides.13Here it was

J. Org. Chem., Vol. 41, No. 24, 1976 3855

Synthesis of a-Dehydrobiotin ineffective. Thus the use of orthophosphoric acid allowed completion of a short synthesis of the antibiotic d-a-dehydrobiotin (1) from the sulfonium salt 7b with all steps going in reasonable yield.

chloride, 0.6 g of sodium acetate, 5 ml of ethanol, and 5 ml of water was heated under reflux for 1h. The reaction mixture was cooled and concentrated in vacuo. The residue was triturated with water and recrystallized from ethanol/hexane to give 0.5 g of 6, m p 202-205 "C dec. Recrystallization from ethanol/hexane gave colorless prisms: mp 202-204 dec; NMR (MeZSO) 6 6.65 (m, 1,-CH=N). Anal. Calcd for C ~ H I ~ N ~ C, O ~44.64; S : H , 6.09; N, 19.52. Found: C, 44.42; H, 6.02; N, 19.25. Dehydration of 3. A mixture of 1 g of 3 , l ml of piperidine, 10 ml of methanol, 10 ml of water, and 1 g ofmalonic acid was allowed to stand at room temperature for 24 h and then warmed on the steam bath for 0.5 h. It was then concentrated in vacuo. The residue was crystallized from water to give an amorphous white solid. Recrystallization from ca. 40 ml of water gave 0.7 g. Recrystallization from methanol/ether gave white needles: mp 200-210 "C dec; MS m / e 364. Anal. Calcd for C16H20N402S2: C, 52.72; H , 5.53; N, 15.37. Found: C, 52.96; H , 5.41; N, 14.55. Treatment of 3 with Acetic Anhydride. A mixture of 1g of 3 and 10 ml of acetic anhydride was heated under reflux for 4 h. The reaction mixture was filtered to remove a small amount of white insoluble material and the filtrate was concentrated in vacuo. The residue was crystallized from ether t o give 0.8 g of product, mp 220-230 "C. Recrystallization from 2-propanol gave fine white needles: m p 258-263 "C dec; MS m / e 448. Anal. Calcd for C20H24N404S2: C, 53.55; H , 5.40; N, 12.50. Found: C, 53.34; H , 5.59; N, 12.34.

"c

Experimental Section Melting points are uncorrected. NMR spectra were recorded on Varian T-60 and HA-100 instruments and are reported in parts per million from internal tetramethylsilane. Infrared and mass spectra were recorded on Perkin-Elmer 137 and CEC-11OB instruments, respectively. Elemental analyses were conducted under the supervision of Dr. F. Scheidl of our microanalytical laboratory. 3-Hydroxy-1,2,3,6,6acr,7,8ac~,8ba-octahydro-5~-pyrido[ 1,2,3-cd]thieno[3,4-d]imidazol-5-one(3). A mixture of 18 ml of 2nitropropane and 65 ml of 3 N sodium hydroxide was heated on the steam bath until the p H was about 7-8. Then 13.2 g of 2c was added and the mixture was heated under reflux for 3.5 h. It was then cooled in an ice bath and the solid collected to give 7.7 g (75%) of product, mp 184-185 "C dec. Recrystallization from water gave colorless bars, 50% recovery: m p 168-178 "C dec; NMR (MezSO) 6 6.65 (s, 1,NH), 5.52 (d, 1,J = 4 Hz, OH), and 5.20 ppm (m, 1,NCHO); ir (KBr) 1590 and 1640 cm-I. Anal. Calcd for C ~ H I ~ N Z OC,~47.97; S : H, 6.04; N, 13.99. Found: C, 47.63; H , 6.04; N, 14.14. Crystallography. Crystals of 3 are monoclinic, space group P21, with a = 6.44 (l), b = 8.39 (l),c = 8.38 (1)A, p = 103.67 (5)",and 2 = 2. The intensity data were measured on an automated diffractometer by a peak-top scan technique (narrow 8-28 scans). Nickel-filtered Cu K a radiation and pulse height discrimination were used. An empirical correction was applied to convert the peak top data to integrated scan data; no absorption correction was made ( p = 29.3 cm-l). The crystal used for data collection was approximately 0.05 X 0.15 X 0.20 mm in size. A total of 643 independent reflections were recorded for 0